CN112613343A - Improved YOLOv 4-based river waste monitoring method - Google Patents
Improved YOLOv 4-based river waste monitoring method Download PDFInfo
- Publication number
- CN112613343A CN112613343A CN202011383306.1A CN202011383306A CN112613343A CN 112613343 A CN112613343 A CN 112613343A CN 202011383306 A CN202011383306 A CN 202011383306A CN 112613343 A CN112613343 A CN 112613343A
- Authority
- CN
- China
- Prior art keywords
- detection
- data
- pictures
- steps
- feature
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 53
- 238000012544 monitoring process Methods 0.000 title claims abstract description 25
- 239000002699 waste material Substances 0.000 title claims description 9
- 238000001514 detection method Methods 0.000 claims abstract description 50
- 238000012545 processing Methods 0.000 claims abstract description 13
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 claims abstract description 11
- 238000012549 training Methods 0.000 claims abstract description 10
- 238000005070 sampling Methods 0.000 claims abstract description 9
- 230000004927 fusion Effects 0.000 claims description 7
- 238000003064 k means clustering Methods 0.000 claims description 3
- 241000196324 Embryophyta Species 0.000 claims description 2
- 239000008267 milk Substances 0.000 claims description 2
- 210000004080 milk Anatomy 0.000 claims description 2
- 235000013336 milk Nutrition 0.000 claims description 2
- 238000010586 diagram Methods 0.000 abstract description 11
- 238000004364 calculation method Methods 0.000 abstract description 7
- 238000013135 deep learning Methods 0.000 abstract description 7
- 208000034699 Vitreous floaters Diseases 0.000 abstract description 6
- 238000013145 classification model Methods 0.000 abstract description 3
- 238000007781 pre-processing Methods 0.000 abstract description 3
- 230000000694 effects Effects 0.000 description 7
- 238000005516 engineering process Methods 0.000 description 4
- 241000207199 Citrus Species 0.000 description 2
- 235000020971 citrus fruits Nutrition 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 238000012935 Averaging Methods 0.000 description 1
- 230000003796 beauty Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 238000011897 real-time detection Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/40—Scenes; Scene-specific elements in video content
- G06V20/41—Higher-level, semantic clustering, classification or understanding of video scenes, e.g. detection, labelling or Markovian modelling of sport events or news items
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
- G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/52—Surveillance or monitoring of activities, e.g. for recognising suspicious objects
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- Evolutionary Biology (AREA)
- Software Systems (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- Molecular Biology (AREA)
- Mathematical Physics (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Health & Medical Sciences (AREA)
- Multimedia (AREA)
- Probability & Statistics with Applications (AREA)
- Image Processing (AREA)
Abstract
The invention relates to the technical field of computer vision, in particular to a river floating object monitoring method based on deep learning. The method comprises the steps of monitoring floaters on the water surface of a river channel from a certain angle and a proper distance by utilizing the existing shooting and recording equipment, carrying out scale clustering processing on all marked frames in a data set after obtaining a video image, preprocessing the video data by adopting a Mosaic data enhancement method, identifying by utilizing a trained classification model, and detecting the floaters in real time. The training and detection network of the embodiment of the invention adopts a YOLOv4 network structure, and the up-sampling mode of the characteristic diagram adopts a bilinear interpolation algorithm, so that the model accuracy is ensured, the calculated amount is reduced, and the calculation speed of equipment is improved. The method has the advantages of high accuracy, high detection speed and high application value.
Description
Technical Field
The invention relates to the technical field of computer vision, in particular to a river floating object monitoring method based on improved YOLOv 4.
Background
Floaters in the river have important effects on the beauty of the river, the water quality of the river and the environmental assessment. Nowadays, river course floating objects are often found on site manually or by means of videos, a large amount of labor and time are consumed, and a convenient and reliable method capable of automatically monitoring river course wastes is needed. With the development of machine vision technology, it is possible to automatically identify and monitor river channel wastes based on machine vision.
At present, many methods for detecting a target by using technologies such as deep learning are available, for example, chinese patent publication No. CN 111709381a discloses a road environment target detection method based on YOLOv3-SPP, and the method has better anti-noise performance and recognition accuracy by using a deep learning technology and an image processing technology, and provides possibility for an intelligent driving system to accurately perceive driving. Chinese patent publication No. CN 111553406a discloses a target detection system, method and terminal based on improved YOLO-V3, the system has a smaller network model, accelerates the target detection speed, enhances the network feature fusion effect, and achieves better detection results. The Chinese patent document with publication number CN111709489A discloses a citrus identification method based on improved YOLOv4, and the method can better identify citrus with smaller individuals by improving a YOLOv4 network model structure, adding an up-sampling module and a detection characteristic diagram sensitive to small targets. Chinese patent publication No. CN110348303A discloses an auxiliary water surface patrol system capable of being carried on unmanned boat and a water surface monitoring method, the method adopts YOLOV3 algorithm for detection and identification, and the calculated speed has a certain difference with YOLOV 4.
Although the method has a good detection effect on a specific target, the river waste has different characteristics and has various interferences, such as: the float is in the shape of a small circle, the float is in the shape of a circle, the float is in the shape of. Therefore, data needs to be preprocessed and monitored by using a proper network and algorithm.
Disclosure of Invention
The invention aims to provide a river floating object monitoring method based on deep learning, which realizes the monitoring of the river water surface by identifying floating objects in a video. In order to achieve the above purpose, the present invention adopts a river floating object monitoring method based on modified YOLOv4, as shown in fig. 1, wherein the training process comprises the following steps:
1) acquiring original video data of river channel water surface monitoring, and extracting image data from the original video data;
2) carrying out scale clustering processing on all marked frames in the data set;
3) video data is preprocessed by adopting a Mosaic data enhancement method;
4) inputting batch image data into a network for forward propagation to obtain a detection result;
5) calculating the loss of the detection result and the labeled value;
6) carrying out back propagation according to the loss value and updating the weight according to the learning rate;
7) and repeating the steps 4), 5) and 6) until the network loss is reduced and tends to converge.
The detection process comprises the following steps:
8) taking an image to be detected as input, and extracting picture characteristics through a backbone;
9) extracting feature maps of different depths in the backbone network;
10) taking the extracted feature maps of multiple scales as input of an FPN structure for feature fusion, wherein the up-sampling mode of the feature maps is a bilinear interpolation algorithm;
11) inputting the FPN fused multi-scale feature map into a PAN structure for strong feature positioning to obtain three feature map detection results with different scales;
12) performing nms processing on all feature map detection results to generate a final result, and marking a detection frame and a detection category in an original input image;
13) and extracting the next frame of picture to be detected, and repeating the steps from 8) to 12) to finish the frame-by-frame detection of the video.
According to the technical scheme, the existing shooting and recording equipment is used for monitoring the floating objects on the water surface of the river from a certain angle and a proper distance, after video images are obtained, all marked frames in a data set are subjected to scale clustering processing, then the video data are preprocessed by adopting a Mosaic data enhancement method, then a trained classification model is used for identification, and the floating objects are detected in real time.
In step 2): carrying out scale clustering processing on all marked frames in the data set; adopting a K-Means clustering algorithm to obtain 9 types of prior frames with different scales, wherein the process comprises the following steps:
2-1) firstly randomly selecting 9 sample points in all labeled group Truth sample points as the centers of the clusters (each sample is a four-dimensional vector);
2-2) respectively calculating the distances from all the other sample points to the 9 centers, wherein each sample point belongs to the center point closest to the sample point;
2-3) selecting a new clustering center in a mode of solving the mean value of four dimensions in the newly divided clusters;
2-4) repeating the steps 2-2) and 2-3) until the new cluster center is not changed from the original cluster center or the variation range is within a specified range.
The distance formula adopts an Euclidean distance formula:
here, X, Y are two sample points, Xi,yiRespectively, the coordinate values of each dimension of the two sample points. The clustering results are sorted into (23,29), (37,34), (26,53), (41,90), (94,40), (61,75), (78,135) and (188,206) according to the area size, so that the sizes of the prior frames are found to be greatly different, and the prior frames are respectively distributed to three feature maps 76 × 76, 38 × 38 and 19 as the prior frames.
In step 3): video data is preprocessed by adopting a Mosaic data enhancement method; the method refers to a CutMix data enhancement mode, Mosaic uses four pictures for splicing, can enrich the background of a detected object, and calculates the data of the four pictures at one time during BN calculation, as shown in figure 2, the flow is as follows:
3-1) randomly selecting four pictures;
3-2) respectively turning over, zooming, color gamut changing and the like on the four pictures, and arranging the pictures in four directions;
3-3) combining the pictures and the frames, and splicing the four pictures into a new picture respectively in the order of upper left, lower right and upper right;
3-4) training with the picture data.
The training and detection network of the present invention adopts a YOLOv4 network structure, as shown in fig. 3. The network structure of YOLOv4 is improved on the basis of YOLOv3, and CSPNet (Cross Stage Partial network) is used in the backbone part of the network, and the CSPNet relieves the problem of needing a large amount of reasoning calculation. The method is realized by dividing a characteristic diagram of a base layer into two parts and then combining the two parts by providing a cross-stage layered structure, and has the advantages of reducing repeated gradient information, reducing calculated amount, being beneficial to improving the calculation speed of equipment and not influencing the accuracy of a model. In order to fully utilize feature information extracted from different layers, the YOLOv4 adopts a network structure of FPN, performs up-sampling from top to bottom on feature maps of different layers obtained by down-sampling an input image, and performs concat splicing with an original feature map on the left side to obtain a new feature map. After the FPN features are combined, a PAN structure is added, the combined bottom-layer feature graph is subjected to convolution downsampling again and is spliced with the feature graph with the same scale in the left-side FPN structure, and finally three output feature graphs with different scales are obtained, as shown in figure 3, the combination aims to convey strong positioning features from bottom to top and enhance the accuracy of model feature extraction.
Step 6): calculating the loss of the detection result and the labeled value; in YOLOv4, the loss function of the predicted bounding box is cioulos, which means that the relative evaluation criteria of center distance and aspect ratio are added on the basis of the IOU, wherein the calculation of the IOU is as follows:
LIOU=1-IOU(Boxpre,Boxgt)
Boxpre,Boxgtthe meaning of the predicted bounding box and the real bounding box is the overlapping area of the two. CIOU is calculated as:
wherein alpha is a positive number, and v is a punishment item for measuring the width and high similarity of the predicted value and the true value. w is agt、hgt、wpre、hpreThe width and height of the real value and the predicted value of the bounding box are respectively. The intermediate term of the loss function is a penalty term for measuring the distance of the central point, wherein rho (-) is the calculation of Euclidean distance, Boxpre_ctrAnd Boxgt_ctrAs the center coordinate, c is the diagonal length of the smallest bounding box of the predicted bounding box and the real bounding box.
Step 10): taking the extracted feature maps of multiple scales as input of an FPN structure for feature fusion; in the present embodiment, bilinear interpolation is used as the upsampling method, as shown in fig. 4. In a conventional nearest neighbor interpolation method, a pixel value of a pixel point to be interpolated is set to be a pixel value closest to 4 surrounding pixel points in an original image, that is, a pixel value most affected by the original image. However, the effect of other pixel points is not considered, and the up-sampling effect is generally not ideal. In order to improve the defects, the method is improved by a bilinear interpolation algorithm.
The bilinear interpolation fully considers the influence of 4 surrounding pixel points, firstly carries out linear interpolation in the horizontal direction to obtain two intermediate pixel values according to the weighted value of the distance between a to-be-sampled point and 4 surrounding adjacent points, and then carries out linear interpolation on the two pixel points obtained by interpolation in the horizontal direction in the vertical direction to obtain a final pixel value, wherein the formula is as follows:
wherein (i, j), (x1, y1), (x1, y2), (x2, y1) and (x2, y2) are the coordinate values of the point to be interpolated and the 4 points around the point to be interpolated A, B, C, D, respectively, and f (i, j), f (a), f (b), f (c) and f (d) are the pixel values of the point to be interpolated and the 4 points around the point to be interpolated A, B, C, D, respectively.
Step 12): performing nms processing on all feature map detection results to generate a final result, and marking a detection frame and a detection category in an original input image; the detection categories comprise 8 types, namely leaves, plastic bags, aquatic weeds, branches, bottles, milk boxes, plastic wastes and spherical wastes.
In the aspect of algorithm detection speed, the frame number of the experimental device before and after model improvement is 82FPS, and the input video is detected frame by frame during detection as shown in FIG. 5, so that the standard of real-time target detection is met.
Compared with the prior art, the invention has the advantages that:
according to the technical scheme, the existing shooting and recording equipment is used for monitoring the floaters on the water surface of the river channel from a certain angle and a proper distance, after video images are obtained, clustering processing of scales is carried out on all marked frames in a data set, preprocessing is carried out on the video data by adopting a Mosaic data enhancement method, then recognition is carried out by using a trained classification model, and real-time detection is carried out on the floaters. The training and detection network of the embodiment of the invention adopts a YOLOv4 network structure, and the up-sampling mode of the characteristic diagram adopts a bilinear interpolation algorithm, so that the model accuracy is ensured, the calculated amount is reduced, and the calculation speed of equipment is improved. The method has the advantages of high accuracy, high detection speed and high application value.
Drawings
Fig. 1 is a flow chart of a river floating object monitoring method based on deep learning in the embodiment of the invention;
FIG. 2 is a diagram illustrating the effect of the Mosaic data enhancement according to the embodiment of the present invention;
FIG. 3 is a diagram of a YOLOv4 network according to an embodiment of the present invention;
FIG. 4 is a schematic diagram of bilinear interpolation according to an embodiment of the present invention;
fig. 5 is a schematic diagram of frame-by-frame detection according to an embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention will be further described with reference to the following embodiments and accompanying drawings. It is to be understood that the embodiments described are only a few embodiments of the present invention, and not all embodiments. All other embodiments, which can be derived by a person skilled in the art from the described embodiments without any inventive step, are within the scope of protection of the invention.
Unless defined otherwise, technical or scientific terms used herein shall have the ordinary meaning as understood by one of ordinary skill in the art to which this invention belongs. The use of the word "comprise" or "comprises", and the like, in the context of this application, is intended to mean that the elements or items listed before that word, in addition to those listed after that word, do not exclude other elements or items. The terms "connected" or "coupled" and the like are not restricted to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "upper", "lower", "left", "right", and the like are used merely to indicate relative positional relationships, and when the absolute position of the object being described is changed, the relative positional relationships may also be changed accordingly.
Examples
Referring to fig. 1 to 5, the present embodiment aims to provide a river floating object monitoring method based on deep learning, which realizes monitoring of river water surface by identifying floating objects in a video. In order to achieve the purpose, the river floating object monitoring method based on the improved YOLOv4 is adopted by the invention and is divided into two processes of network training and detection, as shown in fig. 1. Fig. 1 is a flow chart of a river course floater monitoring method based on deep learning in the embodiment of the invention, wherein the training process comprises the following steps:
the method comprises the steps of preparing, acquiring original video data of river channel water surface monitoring, and extracting image data from the original video data.
And S100, carrying out scale clustering processing on all marked frames in the data set, and obtaining 9 types of prior frames with different scales by adopting a K-Means clustering algorithm.
S101, firstly, randomly selecting 9 sample points in all labeled group Truth sample points as clustering centers (each sample is a four-dimensional vector);
s102, respectively calculating the distances from all the rest sample points to the 9 centers, wherein each sample point belongs to the center point closest to the sample point;
s103, selecting a new clustering center in a mode of averaging in four dimensions from the newly divided clusters;
s104, repeating the steps 102) and 103) until the new clustering center is not changed from the original clustering center.
Step S200, preprocessing the video data by using a Mosaic data enhancement method, as shown in fig. 2. Fig. 2 is a diagram illustrating a Mosaic data enhancement effect according to an embodiment of the present invention.
In step S201, four pictures are randomly selected.
Step S202, the four pictures are respectively turned, zoomed, changed in color gamut and the like, and are well arranged according to four directions. For example, a picture in the data set is flipped clockwise by 90 degrees, 180 degrees and 270 degrees, and the brightness and contrast of the picture are randomly adjusted during flipping, and random scaling is performed to simulate different light conditions and different angles. The image data volume can be expanded, and the acquired data can be fully utilized.
And step S203, combining the pictures and the frames, and splicing the four pictures into a new picture respectively in the order of upper left, lower right and upper right.
Step S204, training is carried out by using the picture data.
Step S300, inputting a batch image data into the network for forward propagation to obtain a detection result.
Step S400, calculating loss of the detection result and the labeled value.
And step S500, performing back propagation according to the loss value, and updating the weight according to the learning rate.
And step S600, repeating the steps S300, S400 and S500 until the network loss is continuously reduced and tends to converge, and at the moment, explaining that the accuracy of the model tends to be stable.
The detection process is shown in fig. 1 and comprises the following steps:
and S700, taking the image to be detected as input, and extracting picture characteristics through a backbone.
And step S800, extracting feature maps of different depths in the backbone network.
And S900, taking the extracted feature maps of multiple scales as input of the FPN structure for feature fusion, wherein the up-sampling mode of the feature maps is a bilinear interpolation algorithm.
And S1000, inputting the multi-scale characteristic diagram after FPN fusion into the PAN structure for strong characteristic positioning to obtain three characteristic diagram detection results with different scales.
Step S1100, performing nms processing on all feature map detection results to generate a final result, and marking a detection frame and a detection category in an original input image.
Step S1200, detecting frame by frame, as shown in fig. 5, repeatedly executes steps S700 to S1100. And then outputting and displaying the result, wherein the detection result is that the position of each frame of detected image is marked by a rectangular frame and the category of each frame of detected image is displayed.
Claims (5)
1. A river course floater monitoring method based on improved YOLOv4 is characterized in that the network training process comprises the following steps:
1) acquiring original video data of river channel water surface monitoring, and extracting image data from the original video data;
2) carrying out scale clustering processing on all marked frames in the data set;
3) video data is preprocessed by adopting a Mosaic data enhancement method;
4) inputting batch image data into a network for forward propagation to obtain a detection result;
5) calculating the loss of the detection result and the labeled value;
6) carrying out back propagation according to the loss value and updating the weight according to the learning rate;
7) and repeating the steps 4), 5) and 6) until the network loss is reduced and tends to converge.
The detection process comprises the following steps:
8) taking an image to be detected as input, and extracting picture characteristics through a backbone;
9) extracting feature maps of different depths in the backbone network;
10) taking the extracted feature maps of multiple scales as input of an FPN structure for feature fusion, wherein the up-sampling mode of the feature maps is a bilinear interpolation algorithm;
11) inputting the FPN fused multi-scale feature map into a PAN structure for strong feature positioning to obtain three feature map detection results with different scales;
12) performing nms processing on all feature map detection results to generate a final result, and marking a detection frame and a detection category in an original input image;
13) and extracting the next frame of picture to be detected, and repeating the steps from 8) to 12) to finish the frame-by-frame detection of the video.
2. The improved YOLOv 4-based river floating object monitoring method according to claim 1, wherein the step 2) of performing scale clustering on all labeled borders in the data set comprises the steps of obtaining 9 types of prior frames with different scales by using a K-Means clustering algorithm, wherein the process comprises the following steps:
2-1) firstly randomly selecting 9 sample points in all labeled group Truth sample points as the centers of the clusters (each sample is a four-dimensional vector);
2-2) respectively calculating the distances from all the other sample points to the 9 centers, wherein each sample point belongs to the center point closest to the sample point;
2-3) selecting a new clustering center in a mode of solving the mean value of four dimensions in the newly divided clusters;
2-4) repeating the steps 2-2) and 2-3) until the new cluster center is not changed from the original cluster center or the variation range is within a specified range.
The distance formula adopts an Euclidean distance formula:
here, X, Y are two sample points, Xi,yiRespectively, the coordinate values of each dimension of the two sample points. The clustering results are sorted into (23,29), (37,34), (26,53), (41,90), (94,40), (61,75), (78,135) and (188,206) according to the area size, so that the sizes of the prior frames are found to be greatly different, and the prior frames are respectively distributed to three feature maps 76 × 76, 38 × 38 and 19 as the prior frames.
3. The method for monitoring the river floating objects based on the improved YOLOv4 as claimed in claim 1, wherein the video data in step 3) is preprocessed by a Mosaic data enhancement method, the method refers to a CutMix data enhancement mode, Mosaic uses four pictures for splicing, the method can enrich the background of the detected object, and the data of the four pictures are calculated at one time when BN is calculated, and the process is as follows:
3-1) randomly selecting four pictures;
3-2) respectively turning over, zooming, color gamut changing and the like on the four pictures, and arranging the pictures in four directions;
3-3) combining the pictures and the frames, and splicing the four pictures into a new picture respectively in the order of upper left, lower right and upper right;
3-4) training with the picture data.
4. The method for monitoring the river course floating objects based on the improved YOLOv4 as claimed in claim 1, wherein the feature maps of multiple scales extracted in step 10) are used as the input of the FPN structure for feature fusion, and the feature map upsampling mode is a bilinear interpolation algorithm:
the bilinear interpolation fully considers the influence of 4 surrounding pixel points, firstly carries out linear interpolation in the horizontal direction to obtain two intermediate pixel values according to the weighted value of the distance between a to-be-sampled point and 4 surrounding adjacent points, and then carries out linear interpolation on the two pixel points obtained by interpolation in the horizontal direction in the vertical direction to obtain a final pixel value, wherein the formula is as follows:
wherein (i, j), (x1, y1), (x1, y2), (x2, y1) and (x2, y2) are the coordinate values of the point to be interpolated and the 4 points around the point to be interpolated A, B, C, D, respectively, and f (i, j), f (a), f (b), f (c) and f (d) are the pixel values of the point to be interpolated and the 4 points around the point to be interpolated A, B, C, D, respectively.
5. The river course floating object monitoring method based on improved YOLOv4 as claimed in claim 1, wherein the step 12) is performed by performing nms processing on all feature map detection results to generate a final result and marking out detection frames and detection categories in an original input image; the detection categories comprise 8 types, namely leaves, plastic bags, aquatic weeds, branches, bottles, milk boxes, plastic wastes and spherical wastes.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202011383306.1A CN112613343B (en) | 2020-12-01 | 2020-12-01 | River waste monitoring method based on improved YOLOv4 |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202011383306.1A CN112613343B (en) | 2020-12-01 | 2020-12-01 | River waste monitoring method based on improved YOLOv4 |
Publications (2)
Publication Number | Publication Date |
---|---|
CN112613343A true CN112613343A (en) | 2021-04-06 |
CN112613343B CN112613343B (en) | 2023-12-15 |
Family
ID=75228450
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202011383306.1A Active CN112613343B (en) | 2020-12-01 | 2020-12-01 | River waste monitoring method based on improved YOLOv4 |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN112613343B (en) |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113191228A (en) * | 2021-04-20 | 2021-07-30 | 上海东普信息科技有限公司 | Express item casting identification method, device, equipment and storage medium |
CN113239863A (en) * | 2021-05-31 | 2021-08-10 | 中冶京诚工程技术有限公司 | River channel monitoring method and device |
CN113627299A (en) * | 2021-07-30 | 2021-11-09 | 广东电网有限责任公司 | Intelligent wire floater identification method and device based on deep learning |
CN113705111A (en) * | 2021-09-22 | 2021-11-26 | 百安居信息技术(上海)有限公司 | Fitment furniture automatic layout method and system based on deep learning |
CN114937195A (en) * | 2022-03-29 | 2022-08-23 | 江苏海洋大学 | Water surface floating object target detection system based on unmanned aerial vehicle aerial photography and improved YOLO v3 |
CN115169855A (en) * | 2022-06-29 | 2022-10-11 | 郑州轻工业大学 | Unsafe state detection method based on digital twin workshop mixed data set |
Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104616247A (en) * | 2015-02-10 | 2015-05-13 | 天津大学 | Method for aerial photography map splicing based on super-pixels and SIFT |
CN109033934A (en) * | 2018-05-25 | 2018-12-18 | 江南大学 | A kind of floating on water surface object detecting method based on YOLOv2 network |
CN111259827A (en) * | 2020-01-19 | 2020-06-09 | 河海大学 | Automatic detection method and device for water surface floating objects for urban river supervision |
AU2020101011A4 (en) * | 2019-06-26 | 2020-07-23 | Zhejiang University | Method for identifying concrete cracks based on yolov3 deep learning model |
CN111488804A (en) * | 2020-03-19 | 2020-08-04 | 山西大学 | Labor insurance product wearing condition detection and identity identification method based on deep learning |
CN111612784A (en) * | 2020-06-01 | 2020-09-01 | 南通大学 | Steel plate surface defect detection method based on classification-first YOLO network |
CN112001339A (en) * | 2020-08-27 | 2020-11-27 | 杭州电子科技大学 | Pedestrian social distance real-time monitoring method based on YOLO v4 |
-
2020
- 2020-12-01 CN CN202011383306.1A patent/CN112613343B/en active Active
Patent Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104616247A (en) * | 2015-02-10 | 2015-05-13 | 天津大学 | Method for aerial photography map splicing based on super-pixels and SIFT |
CN109033934A (en) * | 2018-05-25 | 2018-12-18 | 江南大学 | A kind of floating on water surface object detecting method based on YOLOv2 network |
AU2020101011A4 (en) * | 2019-06-26 | 2020-07-23 | Zhejiang University | Method for identifying concrete cracks based on yolov3 deep learning model |
CN111259827A (en) * | 2020-01-19 | 2020-06-09 | 河海大学 | Automatic detection method and device for water surface floating objects for urban river supervision |
CN111488804A (en) * | 2020-03-19 | 2020-08-04 | 山西大学 | Labor insurance product wearing condition detection and identity identification method based on deep learning |
CN111612784A (en) * | 2020-06-01 | 2020-09-01 | 南通大学 | Steel plate surface defect detection method based on classification-first YOLO network |
CN112001339A (en) * | 2020-08-27 | 2020-11-27 | 杭州电子科技大学 | Pedestrian social distance real-time monitoring method based on YOLO v4 |
Non-Patent Citations (1)
Title |
---|
徐佰祺 等: "联合多尺度特征和注意力机制的遥感影像飞机目标检测", 测绘科学技术学报 * |
Cited By (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113191228A (en) * | 2021-04-20 | 2021-07-30 | 上海东普信息科技有限公司 | Express item casting identification method, device, equipment and storage medium |
CN113239863A (en) * | 2021-05-31 | 2021-08-10 | 中冶京诚工程技术有限公司 | River channel monitoring method and device |
CN113239863B (en) * | 2021-05-31 | 2024-05-14 | 中冶京诚工程技术有限公司 | River channel monitoring method and device |
CN113627299A (en) * | 2021-07-30 | 2021-11-09 | 广东电网有限责任公司 | Intelligent wire floater identification method and device based on deep learning |
CN113627299B (en) * | 2021-07-30 | 2024-04-09 | 广东电网有限责任公司 | Wire floater intelligent recognition method and device based on deep learning |
CN113705111A (en) * | 2021-09-22 | 2021-11-26 | 百安居信息技术(上海)有限公司 | Fitment furniture automatic layout method and system based on deep learning |
CN113705111B (en) * | 2021-09-22 | 2024-04-26 | 百安居网络技术(上海)有限公司 | Automatic layout method and system for decoration furniture based on deep learning |
CN114937195A (en) * | 2022-03-29 | 2022-08-23 | 江苏海洋大学 | Water surface floating object target detection system based on unmanned aerial vehicle aerial photography and improved YOLO v3 |
CN115169855A (en) * | 2022-06-29 | 2022-10-11 | 郑州轻工业大学 | Unsafe state detection method based on digital twin workshop mixed data set |
CN115169855B (en) * | 2022-06-29 | 2023-05-26 | 郑州轻工业大学 | Unsafe state detection method based on digital twin workshop mixed data set |
Also Published As
Publication number | Publication date |
---|---|
CN112613343B (en) | 2023-12-15 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN112613343B (en) | River waste monitoring method based on improved YOLOv4 | |
CN109740465B (en) | Lane line detection algorithm based on example segmentation neural network framework | |
CN109902600B (en) | Road area detection method | |
CN111914838B (en) | License plate recognition method based on text line recognition | |
CN113627228B (en) | Lane line detection method based on key point regression and multi-scale feature fusion | |
CN111563415A (en) | Binocular vision-based three-dimensional target detection system and method | |
CN113158768B (en) | Intelligent vehicle lane line detection method based on ResNeSt and self-attention distillation | |
CN109726627A (en) | A kind of detection method of neural network model training and common ground line | |
CN109145747A (en) | A kind of water surface panoramic picture semantic segmentation method | |
CN114724120B (en) | Vehicle target detection method and system based on radar vision semantic segmentation adaptive fusion | |
CN112434586B (en) | Multi-complex scene target detection method based on domain self-adaptive learning | |
CN113688836A (en) | Real-time road image semantic segmentation method and system based on deep learning | |
CN116051953A (en) | Small target detection method based on selectable convolution kernel network and weighted bidirectional feature pyramid | |
CN113313732A (en) | Forward-looking scene depth estimation method based on self-supervision learning | |
CN113762409A (en) | Unmanned aerial vehicle target detection method based on event camera | |
CN111738071B (en) | Inverse perspective transformation method based on motion change of monocular camera | |
CN113313047A (en) | Lane line detection method and system based on lane structure prior | |
CN112633274A (en) | Sonar image target detection method and device and electronic equipment | |
CN115019201B (en) | Weak and small target detection method based on feature refinement depth network | |
CN112613392A (en) | Lane line detection method, device and system based on semantic segmentation and storage medium | |
CN116091946A (en) | Yolov 5-based unmanned aerial vehicle aerial image target detection method | |
CN116311218A (en) | Noise plant point cloud semantic segmentation method and system based on self-attention feature fusion | |
CN112541943A (en) | Robot positioning method based on visual road signs | |
CN115511827A (en) | Industrial product image angle detection and correction method based on deep learning | |
CN115223112A (en) | Method for detecting to-be-performed area based on semantic segmentation |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |